Minutes of 42nd FFIR meeting on 10/30/2002

The meeting was held in a room of 425 at KEK, 10:00-12:00, 10/30/2002. We discussed on beam delivery system, fast feedback system, collimation system and support tube R&D, etc. .

At beginning of the meeting, a new face was introduced, who is Mr. Ikegami, a graduate student (D1) at Kyoto university (Prof. Iwashita) and he is interested in a R&D of permanent magnet for final quadrupole magnet as well as the optics.

(1) Simulation on beam delivery system (Tanabe, University of Tokyo)

(transparencies, 6 pages, pdf ,304KB ) The simulation program (GEANT4) was improved with help of Prof. Graham Blair (Royal Holloway University) who visited University of Tokyo on this October. Major issues were improvement of accuracy of intersecting geometrical boundaries and a bug-fix in the FieldManager for appropriate transportation at the boundary of the new volume, and to add an option to use analytical stepper. The analytic stepper employs a transfer matrix (linear optics) for bending and quadrupole magnets, and it employs second order linear approximation for sext-,oct-- and dec. magnets. This option significantly saves a CPU time although the performance is as good as that with no option.

With the improvements, beam sizes at IP were calculated to be sigma*x(y) = 214.5 (2.73)nm with +/- 0.3% (Gaussian sigma) energy spread,, while they were calculated to be sigma*x(y) = 2750 (238.7)nm with only bending and quadrupole magnets. The results are in very good agreement with the design values of 211(2.7)nm. In the simulation, no edge effect is considered at all the magnets.

Future study includes an improvement of CPU time, an estimation of synchrotron background, muon production and neutron production. The neutron production and transportation will be investigated in comparison with FLUKA and MARS programs.

(2) Fast feedback system(N.Delerue, KEK)

(transparencies, 10 pages, pdf ,1.3MB ) After short introduction of the fast feedback system, Nicolas explained three models for the system. The first one is "simple model" which has been presented at the previous meeting. Major characteristics of this model is an over-correction with high gain and high luminosity loss at low gain. In order to avoid overshooting of luminosity loss due to the over-correction, a delayed loop was added in the system, which is called as "delayed model".

The delayed model tries to correct the offset "at once" with high gain(~1). So the delayed loop inhibits consecutive feedback-kicks for a round trip delay time which is assumed to be 14nsec (10 bunches) at present. The model works fine at a low initial offset. However, it has long convergence time at higher offset with using the same gain as that of the low offset because of non-linearity in a relation between deflection angle and offset. If the (linear) gain is adjusted at the high offset, corrected offset starts to oscillate. This model also has more sensitive to white noise than the simple one. When the white noise is bunch to bunch jitter with sigma=0.1 sigma*y, the delayed model has a higher luminosity loss of 6% than 1.6% of the simple model.

Improvements have been suggested by S.Smith (SLAC). The non-linearity can be approximately by three linear curves. An improved model can switch gains between low and high deflection angles(offsets) which may depending on noise conditions. Nicolas demonstrated a relatively fast convergence with this model for a case of the initial offset of 10 sigma*y. He also showed a schematic diagram of electronics circuit.

He would like to start a hardware design. In this afternoon, we will discuss on the hardware with Hayano-san in details.

(3) Collimation system(T.Ohgaki, KEK)

(transparencies, 11 pages, pdf ,476KB ) Ohgaki found a bug at a calculation of synchrotron radiations in SAD. After the debug, he made a SAD script of photon.sad to calculate SR photon distribution after SAD tracking. So, this script can calculate the radiation-profile at any position as well as the source positions, emitting angles. Transverse profile was shown at IP, which consists of Gaussian distribution (sigma*x(y)=0.727(0.186)mm together with horizontally asymmetric tail (0 - about 2cm) from bending magnets. A mask should be necessary to shield against the tail. In the Gaussian distribution, there is a sharp peak ( like a delta-function) in both directions (x,y). The peaks must be synchrotron radiations from the nearest focusing magnets, which is to be confirmed.

At the beam delivery system, 4 muon attenuators have been installed in the simulation of MUCARLO (Namito-version), which are 37mx2,45m and 70m long . Two additional muon attenuators of 20m, 25m long could be installed at the upstream. Without the additional ones, the effect was calculated to be 50% increase of e/mu which is the number of collimated electrons to produce one muon at the detector after the MUCARLO-tracking. Filling sand instead of attenuators inside of the tunnel, the e/mu could increase by an order of magnitude. Combination of the attenuators and sand, the e/mu increases by 50% more. If large-size attenuators(inner 18cm/outer 78cm) are used, the e/mu can increase by a factor of 10. Finally, he found that the additional two attenuators are very effective to increase the e_mu, especially at the furthest collimation position (about 1400m from IP), where the e/mu increases in 100 times. The best place of attenuator was also found to be just at back of collimator (absorber). He will optimized their locations. A comparison with the JLC original collimation system, which uses 120m long attenuators at each collimator, must be very interested in.

(4) Support tube R&D (Yamaoka, KEK)

He has ordered to manufacture a prototype support tube of 1/10. During he is waiting for the tube, he will investigate the detailed effect of middle part in the support tube. The middle part is a very thin tube connecting with two heavy (tungsten) ones, which has been found to be very effective to recover the oscillation properties.

The next meeting will be on 20 November (Wed.), 2002 10:00 - 12:00 at 3 gokan, 425.